This preview shows page 1. Sign up to view the full content.
Unformatted text preview: nd that on a lightly loaded machine, programs of almost arbitrary duration can be measured with an error of less than 1%. On a heavily loaded machine, only programs with durations less than around 10 ms can be measured accurately. Practice Problem 9.7:
Suppose we wish to measure a procedure that requires Ø milliseconds. The machine is heavily loaded and hence will not allow our measurement process to run more than 50 ms at a time. A. Each trial involves measuring one execution of the procedure. What is the probability this trial will be allowed to run to completion without being swapped out, assuming it starts at some arbitrary point within the 50 ms time segment? Express your answer as a function of Ø, considering all possible values of Ø. B. What is the expected number of trials required so that three of them are reliable measurements of the procedure, i.e., each runs within a single time segment? Express your answer as a function of Ø. What values do you predict for Ø ¾¼ and Ø ¼? 476 Observations CHAPTER 9. MEASURING PROGRAM EXECUTION TIME These experiments demonstrate that the Ã -best measurement scheme...
View Full Document
- Spring '10
- The American